Self-Assembly of a Pd2L4 Hydrazone Molecular Cage Through Multiple Reaction Pathways

1. Introduction

Molecular cages are preorganized hosts with a central cavity that provides enhanced host–guest properties compared to less preorganized systems such as macrocycles [1,2], aiming to mimic the sophisticated cavity and functions of enzymes [3,4,5,6]. Chemists have developed synthetic methods to prepare both metal-organic cages and purely organic cages resulting in a wide range of structures with size- and shape-dependent host–guest properties [7,8,9]. Encapsulation in the cavity of cages results in different effects on the guest molecule, ranging from activation for catalytic reactions to protection from the surrounding media. These effects yielded multiple applications of molecular cages, including catalysis [10,11,12,13,14,15], sensing of chemicals [16,17,18,19,20,21,22,23], stabilization of chemical species [24,25], separation process [26,27,28], removal of pollutants from water [29,30,31,32] and biological applications [33,34,35,36,37,38,39,40,41,42,43] among many others [7,8,44,45].

The self-assembly of molecular cages from the constituent building block involves in most cases numerous reversible steps. Reversibility is key for error correction of improperly self-assembled by-products produced during the self-assembly process to give the final cage product [46,47,48]. For this, cage building blocks must have a specific shape and geometry that provides an appropriate preorganization in a similar fashion to macrocycles [7,49]. Besides the geometric requirements of the building blocks, both thermodynamic and kinetic are key in order to set up the appropriate reaction conditions, including concentration of reagents, reaction time, and temperature [46,47,48,50]. In this regard, computational modelling has been extensively used to design cages with specific geometries and properties such as predicting host-guest affinity [51,52,53,54,55,56,57].

Typically, metal-organic cages are prepared by the self-assembly reaction of ligands with metals forming metal–ligand bonds [8,58,59], and purely organic cages are prepared by the self-assembly reaction of ligands with complementary reactivity through reversible reactions such as imine and hydrazone bond formation [7,60,61]. There are examples of cages containing both metal–ligand bonds and reversible organic bonds, allowing their assembly via either type of bond, even both bonds simultaneously [8,62,63,64] Focusing in Pd(II) containing cages, the formation of Pd₂L₄ cages involves the self-assembly of two Pd(II) ions and four ditopic ligands, typically containing pyridine moieties [65,66,67,68], though Pd–pyridine coordination bonds [69]. Regarding hydrazone containing cages, hydrazone bonds are both reversible and robust [70], allowing the preparation of cages through the condensation reaction between hydrazide and aldehyde-containing building blocks [34,71,72,73,74,75,76,77,78,79]. Combining both strategies, Crowley and his team showed the feasibility of using simultaneously in the cage self-assembly both Pd–pyridine coordination bonds and hydrazone bonds [63], and we proved that it is possible to self-assemble cages that contain Pd–pyridine bonds through hydrazone bond formation [34]. These results open the way to explore multiple cage self-assembly reaction pathways involving both Pd–pyridine and hydrazone bonds. For this, we propose studying the self-assembly of a Pd₂L₄ cage through various reaction pathways involving the formation of Pd–ligand bonds, hydrazone bonds, or a combination of both.

In this work, we present a comparative study of the self-assembly of a Pd₂L₄ cage containing Pd–pyridine and hydrazone links [34]. The reversible nature of both bond types enables the self-assembly of the cage through the formation of Pd–ligand bonds or hydrazone bonds, utilizing three distinct reaction pathways (Figure 1). As far as we know, this is the first cage formation comparative study involving Pd–ligand and organic bond formation reactions. In particular, we focused on a Pd₂L₄ cage containing four dihydrazone units with a bended geometry and two [PdPy₄]²⁺ units with a C₄ symmetric geometry that serve to cap the cage [34].

2. Results and Discussion

2.1. Self-Assembly Process of the PD₂L₄ Cage

The self-assembly of a Pd₂L₄ cage containing Pd–pyridine and hydrazone bonds can be performed through multiple ways. In this regard, we proposed 3 possible reactions to prepare the Pd₂L₄ cage C1 (Figure 2). The reaction pathway 1 involves the reaction of dihydrazide ligand 1 and the square planar tetrapyridyl Pd²⁺ motif 2 by hydrazone bond formation [34]; the reaction pathway 2 comprises the reaction of dipyridine ligand 3 with Pd²⁺ by Pd–pyridine bond formation; and the reaction pathway 3 involves the reaction between dihydrazide ligand 1, 3-pyridinecarboxaldehyde, and Pd²⁺ involving simultaneously Pd–pyridine and hydrazone bond formation. In all cases, the self-assembly of cage C1 was performed in deuterated DMSO at 25 ^oC and formation was monitored by ¹H NMR spectroscopy. The solvent DMSO was chosen as it enables the complete solution of the reagents, cage, and reaction intermediates. Quantification of the concentration of the species in solution was performed by integration of the corresponding proton signals considering the concentration of the internal standard 1,4-dimethoxybenzene.

Initially we performed the self-assembly reaction of cage C1 through reaction pathway 1. For this, we placed hydrazide 1 and tetrapyridyl–Pd²⁺ 2 in an NMR tube and monitored the evolution of the reaction by acquiring ¹H NMR spectra at different time intervals for a total of 55 hours (Figure 3). The signals corresponding to the starting materials disappeared rapidly with the simultaneous formation of the signals of the Pd₂L₄ cage. In contrast, few signals of the reaction intermediates could be observed in the reaction mixture, probably due to a combination of the formation of a large number of reaction intermediates in fast exchange with unsymmetrical structures containing chemically inequivalent NMR signals, as often observed in cage self-assembly reactions [46,80].

Then we performed the self-assembly reaction of cage C1 through reaction pathway 2. We reacted hydrazide ligand 3 with Pd(NO₃)₂·to form the cage through Pd–pyridine bond formation. The reaction proceeds smoothly with the formation of cage C1, that is the only product observed in the ¹H NMR spectra (Figure 4). In contrast to reaction pathway 1, after 6 minutes, nearly all signals corresponding to the starting materials are absent, and the predominant signals in the spectra correspond to the cage. This highlights the rapid formation of the Pd–pyridine bonds, resulting in quick cage assembly.

Finally, we performed the self-assembly reaction of cage C1 through reaction pathway 3. This reaction is more complex, as it involves the simultaneous formation of 16 bonds, comprising 8 Pd–pyridine bonds and 8 hydrazone bonds. Indeed, this reaction pathway involves more difficulty, for example, the nitrogen atom of hydrazine 1 may result in competition with pyridine for Pd²⁺ coordination that may disturb the cage self-assembly pathway. However, despite this complexity, the reaction yields the expected C1 cage in a clean self-assembly reaction (Figure 5). This self-assembly experiment highlights the feasibility of the simultaneous formation of Pd–pyridine and hydrazone bonds in a cage self-assembly reaction.

After performing the self-assembly reactions, we carried out a quantitative analysis of the integrals of the ¹H NMR signals for the three reaction pathways to evaluate the kinetics of cage formation. For this, the integral of the internal standard 1,4-dimethoxybenzene was taken into account (Figure 6). We observed that the fastest cage formation is for reaction pathway 2, which gives a 65% yield in 6 minutes. In contrast, reaction pathways 1 and 3, only give a 16% and 17% yield in 6 minutes, respectively. These results highlight that the formation of Pd–pyridine bond has a greater rate than the formation of hydrazone bond (reaction pathway 2 vs. reaction pathway 1). A key observation in reaction pathway 3, which involves the formation of 16 bonds through the self-assembly of 12 building blocks and 2 Pd(II) atoms, is that it exhibits similar reaction kinetics to pathway 1, which only requires the formation of 8 bonds from the self-assembly of 6 building blocks. This observation shows that hydrazone bond formation is the rate limiting step of the cage self-assembly process, in contrast to the fast Pd–pyridine bond formation. Focusing on the final cage formation yield at 55 hours of reaction, both reaction pathways 1 and 3 have similar yields in the range 78–79%, whereas reaction pathway 2 has a 73% yield. Despite all that, reaction pathway 2 is the fastest, the final yield is the lowest, highlighting that the final yield does not depend exclusively on the initial reaction rate. As the cage self-assembly yields are less than 100% for the three reaction pathways, in all three cases by-products are formed.

In order to determine if the by-products formed in the three reaction pathways are visible by ¹H NMR, a close examination of the obtained spectra at the end of the reaction was performed. While reaction pathways 1 and 2 produce clean ¹H NMR spectra showing only the signals of cage C1, reaction pathway 3 displays a small set of additional peaks (Figure 7). This is probably due to the formation of asymmetric oligomeric structures possessing a number of chemically distinct NMR signals that are individually at too low a concentration to be observed by ¹H NMR.

To understand the successful cage self-assembly observed in the three different reaction pathways, we additionally carried out molecular mechanics calculations. We performed a conformational search of the building blocks to obtain the most stable conformations. We observed that all conformations of ligands 1 and 3 have a bent configuration (Figure 8c,d), with good complementarity of ligand 1 to the geometry of metal complex 3 (Figure 8e) and ligand 3 to the square planar geometry of Pd(II). Specifically, the rigidity of the core Ph–O–Ph fragment of ligands 1 and 3 provides a key structural element with an average bent angle of 121 ^o (Figure 8c) and 118 ^o (Figure 8d), respectively. These angles match nicely the 130 ^oC average angle the ligand has in the crystal structure of the cage C1·NO₃ (Figure 8a) and the theoretical angle of 120^o of the chemical representation of cage C1 (Figure 8b). This analysis suggests that the successful self-assembly of the ligands into the cage structure is linked to a favorable preorganization of the building blocks, whose geometry aligns well with the cage structure.

3. Materials and Methods

Materials. All chemicals and solvents were obtained from commercial sources and used without further purification unless specified.

NMR Experiments. ¹H spectra were recorded on a Bruker FT-NMR Avance 400 (Ettlingen, Germany) spectrometer at 300K. Chemical shifts (δ) are reported in parts per million (ppm) and referenced to residual solvent peak.

Molecular Modelling. The structure of ligands was modelled with the Spartan’ 20 software, using the built-in conformational search algorithm using the MMFF force field [81].

3.1. Synthesis of Ligands and Cages

Compounds 1, 2, 3, and C1 were prepared as described by our research group as reported in the literature [34].

3.2. Cage Formation Kinetic Experiments

All ¹H NMR kinetic experiments were performed using the following general procedure. To an NMR tube was introduced 1 (2.6 mg, 9.2 µmol) and 2 (3.0 mg, 4.6 µmol) for reaction pathway 1; 3 (4.3 mg, 9.2 µmol) and Pd(NO₃)₂·2H₂O (1.2 mg, 4.6 µmol) for reaction pathway 2; or 1 (2.6 mg, 9.2 µmol), and Pd(NO₃)₂·2H₂O (1.2 mg, 4.6 µmol) for reaction pathway 3. Then a stock solution of 1,4-dimethoxybenzene as internal standard in DMSO-d₆ (600 µL of a 10 mM stock solution) was added for reaction pathways 1 and 2. For reaction pathway 3 a solution of 1,4-dimethoxybenzene as internal standard (600 µL of a 10 mM stock solution) containing 3-pyridinecarboxaldehyde (1.8 µL, 18.4 mmol). The reaction was shaken to get a clear solution of all the components and the crude reaction mixture was monitored by ¹H NMR for 55 hours at 25 ºC. The concentration of all chemical species was determined for each reaction time by the analysis of integrals of the ¹H NMR signals of the cage and the internal standard, reporting the yield as the average. All reactions were performed at least twice, and a representative example is reported in the manuscript.

5. Conclusions

We have performed a study of the self-assembly of a Pd₂L₄ hydrazone molecular cage C1 through 3 different reaction pathways involving formation of Pd–ligand bonds, hydrazone bonds, or a combination of both. Our results show that it is possible to self-assemble the cage structure C1 through 3 different reaction pathways, obtaining yields ranging from 73 to 79%, with the lowest yield observed for pathway 2 (73%) and similar yields for pathways 1 and 3 (78% and 79%, respectively). The fastest initial reaction rate is observed for reaction pathway 2, compared to reaction pathways 1 and 3, that have similar initial reaction rates, indicating that Pd-pyridine bonds are formed faster than hydrazone bonds. In overall, the self-assembly reaction pathway influences the initial reaction kinetics and the final cage yield. We also proved that despite the complexity of the reaction pathway 3, that involves the formation of 16 bonds in contrast to reaction pathways 1 and 2 that only involve the formation of 8 bonds, the cage is formed in a 79% yield. Molecular modelling shows that the ligands have a favorable reorganization, whose geometry matches well with the cage structure. We anticipate that these results will open the way for more complex cage self-assembly designs involving the simultaneous formation of both Pd–ligand and hydrazone bonds.